Non-constant discounting and consumption, portfolio and life insurance rules☆
Introduction
Intertemporal decision problems have been revised in the recent literature since the classical exponential discounting has been found to be unrealistic, at least in some situations, as pointed out in the comprehensive work of Frederick et al. (2002). Hyperbolic discounting has been introduced in order to account for situations where the time preferences of the decision maker change depending on whether these decisions are short-term or long-run actions. In a continuous time setting, several recent papers (e.g., Karp, 2007) have introduced variable rates of time preference. As preferences change continuously over the time horizon, it is usual to refer to the -agent as the decision maker that makes a decision at time in order to differentiate her from the -agent, i.e., the same agent making a decision at a different moment (Strotz, 1956). If the agent is able to precommit her future behavior we obtain the so-called precommitment solution. On the contrary, under the assumption of no precommitment, agents can act in a naive or a sophisticated way. Naive agents make decisions without taking into account that their preferences will change in the future and, consequently, they will be continuously modifying their choices for the future. On the contrary, a sophisticated agent, in the search for a time-consistent policy, will take into account that her future preferences will be different, and she will constrain her current decisions to her future behavior. We refer to Yong (2012) and references therein for a study of the corresponding equilibrium conditions in a stochastic setting.
In this paper we depart from the work of Pliska and Ye (2007) where optimal consumption, portfolio and life insurance rules were studied for an investor with an arbitrary but known distribution of lifetime, and generalize the investor’s time preferences by considering a non-constant discount1 rate within a stochastic framework, extending previous results from Marín-Solano and Navas (2010). Within this setting, we study precommitment, naive and sophisticated solutions for CARA and CRRA utility functions. For some discount functions, we characterize sophisticated solutions as the solution to a system of two ordinary differential equations that can be solved numerically. As in the standard discounting model, when only hyperbolic discounting is considered, our numerical results confirm the qualitative behavior found in the standard discounting model, in contrast to other results in the literature (see discussion in Section 3).
Section snippets
The model
Consider a decision maker with a working life of years and an initial wealth of who earns money at a rate . is a fixed planning horizon that can be interpreted as the retirement time of the wage earner. While she is alive, at every moment the agent must decide how to allocate her money between consumption, investment and the purchase of life insurance. The introduction of a life insurance contract in the model is very relevant given the importance of the bequest in the
Numerical illustrations
For the numerical example we consider as a baseline case a 25 years old agent who earns at time a wage of , and with an expected date of retirement of . Following Milevsky (2006), in order to compute the probability of dying at age we use the density function , where , is the Gompertz law of mortality with parameters and . Besides, the agent can invest her wealth in a risk-free asset yielding a return of
References (11)
Non-constant discounting in continuous time
Journal of Economic Theory
(2007)- et al.
Consumption and portfolio rules for time-inconsistent investors
European Journal of Operational Research
(2010) - et al.
Optimal life insurance purchase and consumption/investment under uncertain lifetime
Journal of Banking & Finance
(2007) - et al.
Time-consistent portfolio management
SIAM Journal of Financial Mathematics
(2012) - et al.
Time discounting and time preference: a critical review
Journal of Economic Literature
(2002)
Cited by (0)
- ☆
The authors acknowledge financial support by MEC (Spain) Grant ECO2010-18015. Preliminary versions of this paper were presented in the 13th International Congress on Insurance: Mathematics and Economics (Istanbul, May 2009) and EUROXXIV (Lisbon, July 2010).